Viewpoint | A Preliminary Study on the Copyright of Pictures Generated by AI-The First Case Sentenced in First Instance in China

Acted for infringement with other people's AI pictures

Defendant: Non-commercial use, not intentional infringement

On February 24 this year, Mr. Li used the open source software Stable Diffusion to generate the picture involved by inputting the prompt words. on February 26, he posted the picture on the social platform with the text "spring breeze brings tenderness. Later, Mr. Li discovered that Ms. Liu's Baijia account number published an article entitled "Love in March, in Peach Blossom" on March 2, which used his pictures involved in the case.

Subsequently, Mr. Li filed a lawsuit with the court, believing that the defendant, Ms. Liu, did not obtain his permission and cut off the signature watermark of his picture, which made the relevant users mistakenly believe that the defendant was the original author, seriously infringing upon his right of signature and the right of information network dissemination, and demanded compensation of 5000 yuan for economic losses and an apology.

The defendant, Ms. Liu, argued that she had retrieved the pictures involved through the Internet and used them for matching pictures of her original poetry works. The specific source of the pictures involved could not be provided, nor could the watermark of the pictures involved be explained, and it was not sure whether the plaintiff enjoyed the relevant rights of the original pictures. In addition, she published mainly her original poems, non-involved pictures, and no commercial use, no infringement intention. If the court finds infringement, she is willing to apologize to the plaintiff. However, the amount of economic compensation claimed by the plaintiff is too high, the market price of AI-generated pictures is very low, and she is seriously ill and unable to compensate.

On May 25, the dispute over the infringement of the right of authorship and the right of information network dissemination was filed in the Beijing Internet Court and a collegial panel was formed in accordance with the law. On August 24, the case was heard in public, and the plaintiff and defendant participated online through the electronic litigation platform. It is understood that this case is my country's first case involving the copyright of "AI Wenshengtu". The trial of the case was broadcast live on CCTV and multiple platforms.

Does "AI Wensheng Figure" constitute a work?

The most controversial focus of the case is whether the picture of "spring breeze brings tenderness" constitutes a work and what type of work does it constitute?

In court, the plaintiff, Mr. Li, believes that using AI to generate pictures is the same as taking pictures with a camera. In the past, in an era when camera technology was not as developed as it is today, photographers enjoyed copyright by manually adjusting various parameters to shoot works. Now, the camera is highly intelligent, and the photographer does not need to manually participate, but directly presses the shutter, and the photos taken in this way also enjoy copyright. In order to realize his idea, he needs to use detailed description words and various parameter adjustments, and finally he can let AI draw the picture he wants. "If you simply let AI paint a'beauty at dusk', it will not achieve the desired effect."

Does the AI picture in question constitute a work? According to the Beijing Internet Court, Article 3 of the the People's Republic of China Copyright Law (hereinafter referred to as the Copyright Law) stipulates: "A work referred to in this Law is an intellectual achievement that is original in the fields of literature, art and science and can be expressed in a certain form." To examine whether the object of copyright claimed by the plaintiff constitutes a work, the following elements should be considered: whether it belongs to the field of literature, art and science, whether it is original, whether it has a certain form of expression, and whether it belongs to intellectual achievement. "In this case, from the appearance of the pictures involved, they are no different from the photos and paintings that people usually see. They obviously belong to the field of art and have a certain form of expression, with elements 1 and 3."

With regard to the elements of "intellectual achievement", "intellectual achievement" refers to the results of intellectual activities. Therefore, the work should reflect the intellectual input of natural persons. The court said that in this case, the plaintiff wanted to draw a close-up of a beautiful woman with a photographic style under the dusk light condition, and then he entered the prompt words in the Stable Diffusion model. The art type of the prompt words was "ultra-realistic photo" and "color photo", the main body was "Japanese idol", and the details of the characters such as skin state, eye and braid color were described in detail, the environment is "exterior scene", "prime time" and "dynamic lighting". The characters are presented in "cool posture" and "looking at the lens". The style is "film texture" and "film simulation". At the same time, relevant parameters are set. According to the initially generated pictures, prompt words are added, parameters are adjusted, and finally a satisfactory picture is selected.

Therefore, the court held that from the time the plaintiff conceived the picture involved to the time when the picture involved was finally selected, the plaintiff made certain intellectual investment in the whole process, such as designing the presentation mode of the characters, selecting the prompt words, arranging the order of the prompt words, setting relevant parameters, and selecting which picture meets the expectation. The pictures involved in the case reflect the plaintiff's intellectual input and have the elements of "intellectual achievements.

As for the copyright issues caused by AI and other cutting-edge technologies, the court held that the camera function of smart phones is becoming more and more powerful and easier to use. However, as long as the photos taken by smart phones reflect the original intellectual input of photographers, they still constitute photographic works and are protected by copyright law. The more technology develops, the smarter the tools, and the less human investment, but this does not affect our continued application of the copyright system to encourage the creation of works.

It is worth noting that the court held that at this stage the generative AI model does not have free will and is not a legal subject. When people use artificial intelligence models to generate pictures, they are still essentially people using tools to create. That is, the entire creative process of intellectual input is human rather than artificial intelligence model.

The plaintiff enjoys the copyright of AI pictures.

Infringement of use without permission

In addition, whether the plaintiff enjoys the copyright of the picture involved in the case? Whether the accused act constitutes an infringement, and whether the defendant should bear legal responsibility? It is also the focus of the court debate in this case.

The court ruled that the plaintiff was the person who directly set up the artificial intelligence model involved in the case according to the needs and finally selected the picture involved. The picture involved in the case was directly generated based on the plaintiff's intellectual input and reflected the plaintiff's personalized expression. Therefore, the plaintiff was the author of the picture involved and enjoyed the copyright of the picture involved. At the same time, the court stressed that although the plaintiff was found to enjoy copyright as the author in this case, according to the principle of good faith and the need to protect the public's right to know, the plaintiff should clearly mark the artificial intelligence technology or model it used.

With regard to the determination of infringement, the court held that in the case, the defendant used the pictures involved in the case as accompanying pictures and published them without permission, so that the public could obtain the pictures involved in the case at the time and place of their choice, which infringed on the plaintiff's right of information network dissemination. As for the right of authorship, the defendant cannot explain the specific source of the accused picture and the relevant circumstances of removing the watermark. It can be concluded that the watermark was removed by the defendant, which infringes the plaintiff's right of authorship and should be held responsible.

On November 27, the Beijing Internet Court made a judgment of first instance, finding that the defendant had infringed upon the plaintiff's right of signature and information network dissemination rights for the pictures involved in the case, and should issue a statement on the social platform to apologize to eliminate the impact, and compensate the plaintiff for economic losses of 500 yuan within seven days of the effective date of the judgment.

Independence refers to whether the work is independently created by the author, that is, there is no plagiarism or copying of other people's works. In this respect, the "creative" process of AI-generated images is completely independent and free of manipulation and intervention by human authors. Thus, the AI-generated picture can be considered to be independently authored.

Creativity refers to whether the work reflects the author's ideas, thoughts or feelings, and has a creative personality. Creativity is another important condition for a work to be protected by copyright law. Although the operating mechanism of AI algorithms is different from the human creative process, AI can generate unique and creative pictures by itself after learning a large number of pictures. This kind of creativity is reflected in the fact that AI-generated pictures have external manifestations that are different from other works, beyond simple copying, plagiarism and imitation. Therefore, AI-generated works can be considered creative.

When understanding independence and creativity, you need to pay attention to the following points:

Independence is relative to other works. As long as the creative process of the AI-generated picture is independent and has not been copied or copied from other works, it can be considered to meet the requirement of independence.

Creativity is a relative concept. Although the creativity of AI-generated pictures may be different from the human creative process, as long as AI-generated pictures have external manifestations that are different from other works and have certain uniqueness and creativity, they can be considered to meet the creative requirements.

Independence and creativity are interrelated. Only when AI-generated images satisfy both independence and creativity can they be regarded as works in the sense of copyright law.

Proving that the AI-generated images were created independently can be done in several ways:

1. Copyright protection: Applying for copyright protection for AI-generated pictures in time can protect their originality. Copyright registration ensures that information such as the time of creation and the author of a work is recorded, providing a basis for later proof of originality.

2. Disclosure of technical process: Disclosure of the technical process of AI-generated pictures is also an important means to prove its originality. By disclosing the technical details of AI painting algorithms, data sets, etc., people can understand how the work was created. The disclosure of the technical process can also attract more professionals to evaluate the work, thereby increasing the credibility of the originality of the work.

Recording the creative process: Recording the creative process of AI-generated pictures is also an important way to prove their originality. During the creative process, the AI painting system generates a large number of intermediate results and sketches. Preservation of these intermediate results and sketches can serve as valid evidence to prove the originality of the work. Recording the creative process can also show the creative ideas and technical characteristics of the AI system, further proving the uniqueness of the work.

Contrast with other works: Comparing AI-generated images with other existing paintings is also a way to prove their originality. By comparing with the existing works, we can analyze the differences in the style, composition and color application of the works. If the AI-generated image is significantly different from the existing work, it can prove its originality.

5. Professional evaluation: It is also a common method to prove the originality of AI-generated images through professional evaluation. Professionals are invited to evaluate the works and analyze their artistic value, innovation and other characteristics. Professional evaluation can provide an in-depth analysis of the work from an artistic point of view, providing authoritative proof of the originality of the work.

When disclosing the technical process of AI-generated pictures, you need to pay attention to the following details:

Ensure that the disclosure of the technical process is legal. Before disclosing the technical process, it is necessary to understand the relevant laws and regulations and intellectual property laws to ensure that the disclosure will not infringe on the rights and interests of others.

2. Describe in detail the technical process of AI generating pictures. This includes details such as the algorithms, datasets, and training processes used, as well as the steps and parameters for generating images. This will give people a better understanding of the technical principles and processes of AI-generated images.

3, the disclosure of technical processes need to protect sensitive information. Some information may involve trade secrets or sensitive data that needs to be appropriately hidden or deleted to protect the business interests of the enterprise and data security.

Consider ethical and moral issues. Disclosure of technical processes requires compliance with ethical and moral norms, non-disclosure of information on ethical and moral issues, and respect for human values and ethical norms.

Ensure that the disclosure of technical processes is credible. When disclosing the technical process, it is necessary to provide true and accurate information without exaggeration or false propaganda, so as to increase the public's trust and recognition of the technology.

It should be pointed out here that since there are a large number of so-called "intelligent AI software" on the market, the differences brought about by different background databases, different training times, different algorithms and different times of each software are very obvious. Even if the same software is controlled by the same operator, due to different input instructions or even different sequences, the final pictures will be different. In a sense, this is also the "originality" characteristic of AI-generated images. However, this also requires the relevant copyright owners to produce a great difference with the original traditional software. Because the original traditional software runs the same command, even at different times and different operators, as long as the same instruction is input on the same software of the same version, the same result is still obtained. Therefore, the huge difference between this and traditional software causes the copyright owner of AI-generated pictures to have a strong timeliness when recording his own creative process. Once it is not recorded in time, it will be difficult to supplement.